JATTA BERBERAT NÉE KURKIJÄRVI
Quantitative Magnetic Resonance Imaging of Native and Repaired Articular Cartilage
An Experimental and Clinical Approach
JOKA KUOPIO 2008
KUOPION YLIOPISTON JULKAISUJA C. LUONNONTIETEET JA YMPÄRISTÖTIETEET 239 KUOPIO UNIVERSITY PUBLICATIONS C. NATURAL AND ENVIRONMENTAL SCIENCES 239
Doctoral dissertation To be presented by permission of the Faculty of Natural and Environmental Sciences of the University of Kuopio for public examination in Auditorium L21, Snellmania building, University of Kuopio on Tuesday 23rd September 2008, at 12 noon
Department of Physics and Institute of Biomedicine, Anatomy University of Kuopio
Tel. +358 40 355 3430 Fax +358 17 163 410
http://www.uku.fi/kirjasto/julkaisutoiminta/julkmyyn.html Series Editors: Professor Pertti Pasanen, Ph.D.
Department of Environmental Science Professor Jari Kaipio, Ph.D.
Department of Physics Author’s address: Department of Physics
University of Kuopio P.O. Box 1627 FI-70211 KUOPIO FINLAND
Tel. +41 79 277 62 32 Fax +358 17 162 585 E-mail: jattaberberat@gmail.com Supervisors: Professor Jukka Jurvelin, Ph.D.
Department of Physics University of Kuopio
Docent Miika Nieminen, Ph.D.
Department of Diagnostic Radiology Oulu University Hospital
Reviewers: Docent Antero Koivula, Ph.D.
Department of Oncology and Radiotherapy Oulu University Hospital
Vládimir Mlynárik, Ph.D.
Laboratory of Functional and Metabolic Imaging Ecole Polytechnique Fédérale de Lausanne Lausanne, Switzerland
Opponent: Assistant Professor Young-Jo Kim, M.D., Ph.D.
Harvard Medical School Boston, MA, USA
ISBN 978-951-27-0977-9 ISBN 978-951-27-1092-8 (PDF) ISSN 1235-0486
Kopijyvä Kuopio 2008 Finland
Berberat née Kurkijärvi, Jatta. Quantitative Magnetic Resonance Imaging of Native and Repaired Articular Cartilage: An Experimental and Clinical Approach. Kuopio Univer- sity Publications C. Natural and Environmental Sciences 239. 2008. 90 p.
ISBN 978-951-27-0977-9 ISBN 978-951-27-1092-8 (PDF) ISSN 1235-0486
ABSTRACT
Degenerative joint diseases, such as osteoarthritis (OA), and trauma–based knee injuries damage the articular cartilage. Consequently, the joint function becomes impaired and severe pain decreases the quality of life of countless individuals. Quantitative magnetic resonance imaging (qMRI) is the only non-invasive technique available which can evalu- ate the structural and compositional properties of articular cartilage. In this study, three quantitative1H NMR relaxation techniques were investigatedin vitroat 9.4 T using hu- man and bovine cartilage andin vivoat 1.5 T in orthopaedic patients.
In the present study,T1 andT2relaxation times and Gd-DTPA2−–enhanced MRI of cartilage (dGEMRIC) techniques were used to assess the ability of these techniques to probe the structural and mechanical properties across the cadaver human knee joint, and the results were compared with histological reference techniques (i.e. polarized light microscopy (PLM) and optical density (OD) microscopy) and mechanical testing. The reproducibility of T2 measurementin vitrowas examined in bovine articular cartilage.
The structure of cartilage was studied withT2in the presence of Gd-DTPA2−. The abi- lity of nativeT1to reflect tissue hydration was evaluated. Finally, the ability ofT2and dGEMRIC to evaluatein vivoregeneration of cartilage tissue after autologous chondro- cyte transplantation (ACT) was assessed in patients at 10–15 months after surgery.
MRI techniques reproduced satisfactorily the variations in mechanical properties in human tissue. The mean thicknesses of the different cartilage zones were consistent when determined fromT2and PLM profiles. The characteristic laminar cartilage structure, as detected consistently with PLM andT2, was preserved in the presence of Gd-DTPA2−. The reproducibility ofT2measurements was good. T1relaxation rate displayed a high linear association with the cartilage water content. ACT grafts showed a general trend towards longerT2values for bulk tissue as well as for the superficial and deep tissue as compared to the adjacent native cartilage. dGEMRIC detected no significant differences between the native cartilage and graft tissue.
The present results demonstrate that qMRI can serve as a biomarker for structural (collagen network architecture), compositional (i.e.proteoglycan and water content) and mechanical (compressive stiffness) properties of articular cartilage. Further, by com- bining different quantitative magnetic resonance imaging techniques it is possible to achieve a comprehensive characterization of native cartilage and cartilage repair.
National Library of Medicine Classification: WE 103, WE 300, WE 304, WE 348, WN 185 Medical Subject Headings: Joint Diseases/diagnosis; Osteoarthritis/diagnosis; Cartilage, Articular; Magnetic Resonance Imaging; Contrast Media; Gadolinium DTPA; Biomecha- nics; Histology; Microscopy, Polarization; Transplantation, Autologous; Chondrocytes;
Collagen; Proteoglycans
To my dearest Tarcis, Janina and Olivia
ACKNOWLEDGEMENTS
This work was carried out in the Department of Physics, University of Kuopio, the Ins- titute of Biomedicine, the Department of Anatomy, University of Kuopio and the De- partment of Biomedical NMR, A. I. Virtanen Institute, University of Kuopio. I wish to express my deepest gratitude to everyone who has contributed to this thesis and helped me throughout the study. Especially, I wish to mention the following persons.
First, I would like to issue my warm thanks to my first supervisor Professor Jukka Jurvelin, Ph.D. With his enthusiasm, support and professional supervision, he has guided me through this thesis. It has been a privilege to work in his group.
I owe my deepest gratitude to my second supervisor, Docent Miika Nieminen, Ph.D.
Without his support, encouragement and professional guidance, this work would not exist. It has been very enlightening to work under his supervision.
This study would not have been possible without the help of Mikko Nissi, Ph.D. I wish to thank him for his friendship, comments and technical assistance he has offered to improve this study.
I wish to express my sincere thanks to Professor Heikki Helminen, M.D., Ph.D., for the opportunity to carry out the experiments in the Department of Anatomy. I thank Professor Ilkka Kiviranta, M.D., Ph.D., Anna Vasara, M.D., Ph.D., Risto Ojala M.D., Ph.D.
and Lauri Mattila, M.D. for their support and medical expertise they shared with me during this work. I also wish to thank Professor Mikko Lammi, Ph.D. for his help and quidance on biochemistry used in this study.
I thank the official reviewers of this thesis, Vladimir Mlynárik, Ph.D., and Antero Koivula, Ph.D., for their comments and constructive criticism. I am also grateful to Ewen MacDonald, D.Pharm., for the linguistic review.
I owe many thanks to my friends and collaborators, who have helped and partici- pated in this study. I want to thank the people working in the Institute of Biomedicine, Department of Anatomy in Kuopio, especially Mrs. Eija Rahunen and Mr. Kari Kotikum- pu. I also want to express my gratitude to our BBC-group researchers Eveliina Lammen- tausta, Ph.D., Juha Töyräs, Ph.D., Mikko Laasanen, Ph.D., Simo Saarakkala, Ph.D. and Jarno Rieppo, M.D. for their ideas and cheerful support. Panu Kiviranta, M.D. has been wonderful in helping me to find missing references, and for the great spirit of the BBC- group I would like to thank Rami Korhonen, Ph.D., Mikko Hakulinen, Ph.D., Heikki Nieminen, Ph.D., Jani Hirvonen, M.Sc., Petro Julkunen, Ph.D., Hanna Isaksson, Ph.D., Erna Kaleva, M.Sc., Antti Aula, M.Sc., Janne Karjalainen, M.Sc., Ossi Riekkinen, M.Sc., Tuomo Silvast, M.Sc., Matti Timonen, B.Sc. and Pauno Lötjönen, B.Sc. The fellowship in this group has been something really special and the atmosphere has given me strength and joy to complete this work. I also want to thank the co-authors for their contributions to this study. I owe my thanks to Keski-Suomen Magneetti Oy, Jyväskylä, Finland, and their personnel for contributing data for thein vivopart of the thesis, as well as Atria Lihakunta Oyj, Kuopio, Finland, for providing bovine joints as research material. I also want to address my gratitude to Department of Biomedical NMR, A. I. Virtanen Institute, for providing the NMR facilities for the studies, especially the following persons: Profes- sor Risto Kauppinen, M.D., Ph.D., Docent Olli Gröhn, Ph.D. and Johanna Närväinen, Ph.D. I would also like to thank Docent Juhana Hakumäki, M.D., Ph.D. for his guidance and support on this project.
I am grateful to my parents, Mikko and Pirjo, for their encouragement and conti- nuous support of this work. I am also grateful to my brothers, Antti and Timo, whose cheerful spirit and encouragement shone into this project. I want to thank my daughters,
project. His love made this work possible. Last, I want to thank ourau pair, Johanna, for taking such kind care of my little girls so I could concentrate on the thesis.
Financial support from Kuopio University Hospital (EVO 5031329); Jyväskylä Cen- tral Hospital (EVO BO204); Academy of Finland (grant 205886), National Graduate School of Musculoskeletal Disorders and Biomaterials, Kuopio University Foundation, the Fin- nish Academy of Science and Letters, Paulo Research Foundation and Finnish Cultural Foundation of Northern Savo is gratefully acknowledged.
Rothrist, Switzerland, September, 2008
Jatta Berberat
ABBREVIATIONS AND NOMENCLATURE
ACT autologous chondrocyte transplantation ADC apparent diffusion constant
BF optical birefringence of polarized light CV coefficient of variation
dGEMRIC delayed Gadolinium Enhanced MRI of Cartilage DTI diffusion tensor imaging
ECM extracellular matrix
EDTA ethylenediaminetetraacetic acid
ETL echo train length
FCD fixed charge density
FT Fourier transform
FG femoral groove
FID free induction decay FLC lateral condyle of femur FMC medial condyle of femur
FOV field of view
FSE fast spin echo sequence
GAG glycosaminoglycans
Gd–DTPA2− gadolinium diethylene triamine pentaacetic acid [Gd–DTPA]b molar concentration of Gd–DTPA in bath
[Gd–DTPA]t molar concentration of Gd–DTPA in tissue LPG lateral patellar groove
MACT matrix–associated autologous chondrocyte transplantation
MR magnetic resonance
MRI magnetic resonance imaging
MT magnetization transfer
n number of samples
NMR nuclear magnetic resonance
OA osteoarthritis
OCD osteochondritis dissecans
OD optical density microscopy of Safranin–O stained PGs p statistical significance
PAT patella
PBS phosphate saline buffer
PD proton density
PG proteoglycan
PLM polarized light microscopy
qMRI quantitative magnetic resonance imaging
SI signal intensity
TE echo time
TI inversion time
TLP lateral tibial plateau TMP medial tibial plateau
TR repetition time
aF optical constant
A atomic mass number
Al absorbance
bF optical constant
B magnetic field strength
Bext external magnetic field Bz magnetic field in z–direction B0 static magnetic field strength B1 rf–pulse field strength
E energy between nuclei states
Eeq Young’s modulus
f0 Larmor frequency
Gd dynamic modulus
G1 storage modulus
G2 loss modulus
HA aggregate modulus
H2O water content
I intensity
I0 initial intensity
Ms quantum number corresponding to angular momentum operator
Mx magnetization components along x–axis Mxy magnetization components along xy–plane My magnetization components along y–axis Mz magnetization components along z–axis M0 equilibrium magnetization vector
k Boltzmann’s constant
l optical path length
N number of nuclei in the spin orientation r Pearson correlation coefficient
R relaxivity
R1 T1 relaxation rate R2 T2 relaxation rate
s spin
t time
T1 spin–lattice relaxation time
T1Gd spin–lattice relaxation time in presence of Gd–DTPA2−
T1ρ spin–lattice relaxation time in rotating frame
T2Gd spin–spin relaxation time in presence of Gd–DTPA2−
Tl transmittance
T2 spin–spin relaxation time
T20 component of T2 relaxation time induced by field inhomo- geneities
T2∗ total spin–spin relaxation time α rotation angle of polarized light
γ gyromagnetic ratio
axial strain
l lateral strain
λ wavelength
µ magnetic dipole moment
ν Poisson’s ratio
σ stress
σd dynamic stress
τc correlation time
ω0 angular Larmor frequency
~ Planck’s constant
Z number of protons in a nucleus
LIST OF ORIGINAL PUBLICATIONS
This thesis is based on the following original articles referred by their Ro- man numerals:
I J. E. Kurkijärvi, M. J. Nissi, I. Kiviranta, J. S. Jurvelin and M. T. Niem- inen. Delayed Gadolinium-Enhanced MRI of Cartilage (dGEMRIC) and T2 Characteristics of Human Knee Articular Cartilage: Topo- graphical Variation and Relationships to Mechanical Properties. Magn Reson Med52:41-46, 2004.
II J. E. Kurkijärvi, M. J. Nissi, J. Rieppo, J. Töyräs, I. Kiviranta, M. T.
Nieminen and J. S. Jurvelin. The zonal architecture of human ar- ticular cartilage described by T2 relaxation time in the presence of Gd-DTPA2−. Magn Reson Imaging26:602-607, 2008.
III J. E. Berberat, M. J. Nissi, J. S. Jurvelin and M. T. Nieminen. As- sessment of Interstitial Water Content of Articular Cartilage withT1
Relaxation.submitted to Magn Reson Imaging.
IV J. E. Kurkijärvi, L. Mattila, R. O. Ojala, A. Vasara, J. S. Jurvelin, I.
Kiviranta and M. T. Nieminen. Evaluation of cartilage repair in the distal femur after autologous chondrocyte transplantation usingT2
relaxation time and dGEMRIC.Osteoarthritis and Cartilage15:372-378, 2007.
The original articles have been reproduced with permission of the copy- right holders. The thesis contains also previously unpublished data.
C
ONTENTS1 Introduction 17
2 Articular cartilage 19
2.1 Structure and composition . . . 19
2.2 Mechanical properties of articular cartilage . . . 22
2.3 Osteoarthritis . . . 25
2.4 Cartilage repair . . . 27
3 MRI of articular cartilage 29 3.1 Nuclear magnetic resonance (NMR) . . . 29
3.2 Relaxation . . . 30
3.2.1 Spin–lattice relaxation (T1) . . . 31
3.2.2 Spin–spin relaxation (T2) . . . 32
3.3 dGEMRIC . . . 34
3.4 Other quantitative MRI methods . . . 35
4 Aims of the present study 39 5 Materials and methods 41 5.1 In vitroexperiments . . . 41
5.1.1 Sample preparation . . . 41
5.1.2 Mechanical measurements . . . 43
5.1.3 MRI measurements . . . 44
5.1.4 Polarized light microscopy . . . 45
5.1.5 Measurement of PG content . . . 46
5.1.6 Measurement of water content . . . 47
5.2 In vivoexperiments . . . 47
5.2.1 ACT patients . . . 47
5.2.2 MRI measurements . . . 48
6.1 Topographical variation of cartilage properties . . . 53
6.2 T2 in the presence of Gd-DTPA2− . . . 54
6.3 NativeT1in cartilage . . . 55
6.4 Evaluation of cartilage repair . . . 57
7 Discussion 61 7.1 MRI and structural/functional properties of native cartilage . . . . 61
7.2 MRI and cartilage repair . . . 64
8 Conclusions 67
References 69
Appendix: Original publications 91
C
HAPTERI
Introduction
Osteoarthritis (OA) causes pain and functional disability in countless in- dividuals. OA disrupts the cartilage tissue, increases the water content in the tissue and thickens the subchondral bone [27]. These degenerative changes reduce cartilage stiffness, and impair the mechanical function of the joint [4].
Cartilage lacks the capacity to repair and heal the damage spontane- ously. Therefore, early diagnosis of cartilage degradation is important in order to slow down OA progression and to reduce pain and other symp- toms of the patients with suitable treatments. In trauma–based acute car- tilage injuries, but not in OA patients, one treatment method is surgical cartilage repair.
Conventional methods, such as X–ray imaging of the joint and arth- roscopy, are too insensitive to detect the early OA changes and the latter is invasive. Clinically, a non–invasive diagnostic technique is desirable that would permit the differentiation between different stages of cartilage generation with reproducible results. Furthermore, the ideal technique would be quick to perform and be inexpensive.
Macromolecular changes in tissues are reflected in the magnetic pro- perties of water protons, and therefore magnetic resonance imaging (MRI) represents a non-invasive method to detect early changes in articular carti- lage. In particular,1H nuclear magnetic resonance (NMR) relaxation pro- perties provide multifaceted information about cartilage structure, com- position and function [13, 64, 107, 134, 135, 189, 198].
In this thesis, high field MRI was used to extend the previousin vitro cartilage research, particularly aiming to apply and test quantitative MRI techniques in human articular cartilage and to elucidate variations in the properties of native and repaired human articular cartilage. Furthermore, different MRI techniques were used to assess their ability to probe the
structural and mechanical properties of human cartilage. The possibili- ty to merge different quantitative MRI procedures into one contrast agent imaging session was also evaluated. The water content of articular carti- lage was estimated by MRI. Finally, quantitative MRI methods were app- liedin vivoin patients undergoing cartilage repair.
C
HAPTERII
Articular cartilage
2.1 Structure and composition
Our knee joints are subjected to very high mechanical loads, up to ten times one’s own body weight [129]. As the human knee joint may be ex- posed to one million cycles of loading per year [129], a complex interplay between structural, compositional and mechanical properties of cartilage tissue is needed to carry out this demanding mechanical task. This rep- resents the main role in reducing the mechanical friction and minimizing stresses occurring during joint motion [28].
Articular cartilage, i.e. hyaline cartilage, has a smooth and glistening white appearance. There are no blood vessels or nerves in articular car- tilage. Articular cartilage contains a solid matrix (i.e. chondrocytes, col- lagen and proteoglycan (PG)) and fluid (i.e. interstitial water and elect- rolytes) (figure 2.1, table 2.1). Chondrocytes are metabolically active and synthesize, organize and degrade matrix components. The collagen fibers resist tensile stresses, bind PGs and limit tissue swelling. About 90% of the collagen in articular cartilage is of type II, organized in a triple helix structure of three polypeptide chains [170]. These fibers have poor stretch- ing properties, but the bending ability of collagen is excellent. Collagen fibers supplement the support structure of articular cartilage, and PGs and chondrocytes are bonded within the three–dimensional collagen matrix.
Proteoglycan chains form large aggregate molecules by linking them- selves with hyaluronic acid through the link protein [130]. This aggrecan molecule contains laterally attached, electronegatively charged chondroi- tin- and keratan–sulfated glycosaminoglycan (GAG) chains. They attract water and sodium ions, causing an osmotic pressure that keeps the carti- lage structure stable [66]. The charged groups, referred to as a fixed charge
density (FCD), create also strong repulsive forces against each other [129].
The amount of interstitial water depends mainly on FCD (affected by PGs), swelling pressure (collagen–PG), the organization of collagen net- work and the mechanical strength of the collagen–PG solid matrix (table 2.2) [129]. A major fraction of the water is free to move, a small amount is contained in chondrocytes while 60-70% of water exists around proteo- glycan aggregates [114, 130].
Articular cartilage is structurally inhomogeneous and can be divided into four different zones (figure 2.1). In the superficial zone, collagen fibers run parallel to the cartilage surface and the chondrocytes are flattened, water concentration is at its highest level and the proteoglycan content at its lowest level. In the intermediate zone, collagen fibers are randomly organized and the chondrocytes are round in shape. The water content and the chondrocyte concentration are lower than in the superficial zone.
The proteoglycan content is higher than in the superficial zone. In the deep zone, collagen fibers run perpendicular to the articular surface and the chondrocytes are round and densely packed. The water content is its lowest and the proteoglycan content at its highest in this zone. Finally, the calcified cartilage (below the tidemark) links the cartilage to the sub- chondral bone. Hypertrophic chondrocytes are found in the calcified zone [25, 28, 129, 148].
The thickness of adult human cartilage typically varies between 2–5 mm, the thickness of the calcified cartilage and the subchondral plate in adult humans is ≈ 0.13 mm and ≈ 0.19 mm, respectively [82]. Further- more, the structural architecture of articular cartilage is created by a syn- chronized process of tissue resorption and neoformation [81].
Table 2.1: Structure and content from the main components of articular car- tilage [129] (↓decrease,↑increase).
water collagen proteoglycan
wet weight 60–85% 15–22% 4–7%
dry weight – 50–80% 5–10%
diameter – 20–200nm –
length – – 10−8−10−6m
content in zones surface
l ↓ ↓ ↑
deep
2.1 Structure and composition 21
Table2.2:Macromolecularinteractionsinarticularcartilage[129]. collagen–collagenPG–PGcollagen–PG Type•covalentcross-link•Repulsiveforcesbetween•non-covalentbonding negativelychargedGAGs1)electrostatical 2)mechanical Functioncollagennetwork•compressivestiffness•PGs(-)interactingwith •stiffness•Donnanosmoticpressurecollagen(+) •strength•retainingthemoleculeson•hyaluronatesfrom thetissueaggregatesinteractswith collagenII •swellingpressure
Chondrocytes Superficial zone(10-20%)
Middle zone(40-60%)
Deep zone(30%) Calcified zone Subchondral zone
Interstitial water
Collagen II
Proteoglycan
GAG Hyaluronan acid
Link protein
Proteoglycan aggregate
Tidemark
Figure 2.1: Structure of articular cartilage.
2.2 Mechanical properties of articular cartilage
Cartilage is an inhomogeneous, poroelastic material with nonlinear mec- hanical properties. The mechanical properties vary between species [8, 166], cartilage layers [186] and anatomical locations [86, 110]. Further- more, the measurement direction affects the mechanical response, i.e. the mechanical properties are anisotropic [96]. The mechanical properties can be determined from the load–deformation behavior of the tissue. When
2.2 Mechanical properties of articular cartilage 23 the cartilage is loaded, flow of the interstitial fluid through the extracellu- lar matrix creates a poroviscoelastic response. When the load is removed, the tissue restores its thickness and shape by resorption of the fluid. There is no fluid flow at mechanical equilibrium, and the load is controlled by the solid matrix. Duringinstantaneous loading, elastic deformation without interstitial fluid flow takes place.
It has been shown that the collagen network is mainly responsible for the dynamic response under compression [98, 129], whereas PGs are res- ponsible for the static compressive stiffness of cartilage [95, 129]. Due to tissue complexity, the nonlinear behavior of articular cartilage is most of- ten numerically modeled by using finite–element analysis [46, 194].
Unconfined compression, confined compression and indentation measure- ments are used to study experimentally the mechanical properties of ar- ticular cartilage. Further, three experimental loading techniques are tradi- tionally used: stress–relaxation, creep and dynamic loading (figure 2.2). In unconfined compression, cartilage without the subchondral bone is com- pressed between two smooth and rigid plates, and fluid flows only in the lateral direction. The stiff collagen structure in the superficial tissue allows a lesser lateral expansion as compared to the deeper parts of the cartilage. Inconfined compression, a sample with or without bone is placed in a rigid chamber and compressed with a porous filter. Fluid can only flow axially through the tissue surface into the filter. Under unloaded state, the swelling of PGs is limited by the elastic forces of the collagen network [121]. In the axial direction, compressive stiffness increases [186]
and tensile stress decreases [2] towards the deep cartilage. Unconfined and confined compression measurements can be conducted only in thein vitrosetting.
In unconfined compression, at equilibrium, Young’s modulus of isotropic elastic material is given by
Eeq= σ
, (2.1)
whereσis axial stress andis axial strain in the solid matrix. In confined compression, the equilibrium modulus is called the aggregate modulus HA, related toEandν:
HA= (1−ν)
(1 +ν)(1−2ν)E, (2.2)
where E is Young’s modulus and ν is Poisson’s ratio. The complex dy-
Stress
Stress Stress
Strain Strain
Strain
Time
Time Time Time
Time Time
A B
C
t0
t0 t0
t0
t0
t0
Figure 2.2: (A) In a stress-relaxation experiment, the load (stress) is mea- sured under a constant deformation (strain). (B) In a creep experiment, car- tilage deformation (strain) is measured under constant load (stress). (C) In a dynamic loading experiment, the response of the tissue to cyclic deforma- tion is measured.
namic modulus of cartilage is given by
|Gd|= q
G21 +G22 = σd
, (2.3)
where σd is the dynamic stress, G1 the storage modulus, proportional to the elastically stored energy, and G2 is the loss modulus, i.e. the viscous energy lost in the loading process.
Indentation measurements can be conducted bothin vitroand in vivo.
Cartilage is compressed using an indenter, leading to fluid flow, both in a- xial and lateral directions. In this geometry, the thickness of the superficial zone and the transverse stiffness both play important roles in determining the mechanical response [97].
Typical values for Young’s modulus or aggregate modulus of the intact articular cartilage are 0.2–1.5 MPa [4, 7, 30, 36, 86, 90, 96, 98, 129, 196]. Pois- son’s ratio varies between 0.00–0.43 [8, 36, 87, 96] and values for instanta- neous or dynamic (t →0) modulus are around 1.5–20 MPa [30, 52, 98, 103, 149]. The values depend significantly on the species and the anatomical location of the tissue [8, 86, 110, 166].
2.3 Osteoarthritis 25 Several theoretical models have been devised to predict the mechanical behavior of cartilage. Finite–element analysis (FE) offers the most realis- tic model for the nonlinear behaviour of articular cartilage. Asingle phasic elastic model assumes the material to be inhomogenous and elastic with constant mechanical properties,i.e.isotropic elastic material with uniform mechanical properties in all directions. In this model, only Young’s mo- dulus and Poisson’s ratio are required to characterize the mechanical be- havior of the material [75]. However, the complex nonlinear behavior of cartilage, structural and mechanical anisotropy make the modeling more challenging. The single phasic model is not very realistic, as cartilage con- sists of two phases.
The biphasic modelincludes the two phases of cartilage, i.e. solid and fluid. This is a more realistic presentation of cartilage since the motion of the interstitial fluid has a major impact on the viscoelastic behavior of ar- ticular cartilage. The linear isotropic biphasic model assumes the solid and fluid phases to be incompressible, the solid matrix being isotropic, ho- mogenous and porous whereas the fluid phase is inviscid [128]. This is the most traditional model for characteristing the mechanical behaviour of articular cartilage. However, it fails to predict tensile behaviour [98] or short–term compressive behaviour in unconfined compression [37].
Thetransversely isotropic modeltakes into account the dynamic response better than the isotropic biphasic model [97]. This model can utilize one [195] to two [32] permeability coefficients and five elastic constants. Even though this model includes the parallel oriented collagen fibers in the superficial cartilage layer, it still fails to predict the compression–tension nonlinearity of the tissue.
In thefibril-reinforced poroelastic model, the role of the collagen fibers is considered to provide the stiffness in tension only. This takes into account the compression–tension nonlinearity and time–dependent deformation of the intrinsic viscoelastic matrix [112]. A biphasic model, where an ion phase is included, is called thetriphasic model [171]. This includes tissue swelling, and in this model the stress, strain, ion concentrations, electric fields and flow fields can be defined [68–70].
2.3 Osteoarthritis
Osteoarthritis (OA) is a degenerative joint disease that in its more severe form disrupts the extracellular matrix. Mainly, it appears in patients over the age of 50 years, with the majority being women. In Finland, the ’Health 2000’ study revealed that 16% of men and 32% of women are suffering from knee arthrosis at the age of 75–84 years [6]. In the United States,
approximately 135 000 procedures are done yearly to repair knee defects or to undertake total knee replacements [24]. Thus, OA is a significant health and economic burden to our society [199].
Cartilage has a poor spontaneous capacity to heal itself after an injury or disease [27]. The early stages of OA are often asymptomatic but later the clinical symptoms of OA include limitations of joint movements, de- formation and effusion of the joint, pain and abnormal sounds from the joint (i.e. crepitation with motion). OA can be initiated by an injury, or it may arise after infection (i.e. osteoarthritis) or after metabolic and neuro- logical disorders. OA can also occur spontaneously without any obvious reasons [27].
The earliest symptoms of OA, characteristics of early degeneration, in- clude fibrillation of the collagen fibers, which starts on the articular sur- face and proceeds to the middle zone. There is also loss of PG aggregans, tidemark damage, the appearance of blood vessels from the subchondral bone and subchondral bone modifications [10, 25]. In X–ray images, the first sign are ostephytes. These changes lead to an increase in tissue per- meability, water content and swelling of cartilage [113]. Concomitantly, the mechanical stiffness of articular cartilage becomes reduced. Nonethe- less, this stage is normally asymptomatic for the individual, the cartilage surface seems still regularly glossy and no damage on the surface may be seen. Later, in advanced degeneration, due the changes in osmolarity and ionic charge, chondrocytes release mediators, stimulating metabolic response of cells, aiming to heal the cartilage. An expansion of aggregates and an increase in the water content occurs in the cartilage tissue and also the density of subchondral bone increases. At this stage, changes in glossi- ness and in the color of the cartilage surface may become apparent. Deep defects down to the subchondral bone can be seen [27]. In the third and fi- nal stage of OA,late degeneration, cartilage is fully destroyed, leaving only the thickened and dense subchondral bone for joint contact. The shape of the joint may change and other clinical symptoms, i.e.limping, pain, joint deformity and instability are typically encountered. The loss of cartilage leads to secondary changes in synovial tissue, ligaments, capsules, and joint muscles [27]. Decreased range of motion and muscle atrophy follow after reduced use of the joint.
The diagnostics of OA is usually carried out with X–ray imaging where joint space narrowing, subchondral bone sclerosis, cysts and ostephytes, i.e. typical symptoms of degenerated cartilage, can be observed. Unfor- tunately, the early stage of OA cannot been detected by X–ray evaluation.
Another common diagnostic method is arthroscopic examination. One major prolem of arthroscopy is its invasive nature. Furthermore, it is a
2.4 Cartilage repair 27 subjective examination and depends critically on the individual who per- forms the examination [23]. Although surface fibrillation and the general condition of the joint can be visually detected, the internal cartilage struc- ture and modifications of the subchondral bone cannot be diagnosed.
Magnetic resonance imaging (MRI) is a widely used non–invasive met- hod to examine articular cartilage [62, 126]. The water content of the nor- mal healthy articular cartilage tissue is 60-85% [129]. Since the early OA changes include an increase of the cartilage water content, MRI represents a potential method to detect the early stage of OA. Alterations on the sub- chondral bone may also be diagnosed [44, 190].
2.4 Cartilage repair
The only treatment for advanced osteoarthritis is prosthetic joint replace- ment. However, several methods have been introduced to attempt treat- ment of locally damaged cartilage. Unfortunately, none of these methods have been shown to produce hyaline tissue with a similar composition, structure or mechanical properties as native articular cartilage.
Drilling of the subchondral bone plate to stimulate the cells from the bone marrow to transform them into cartilage cells, was the first carti- lage repair technique in humans which achieved satisfactory results [42].
Later, the drilling was changed to microfracture technique, which today is the most common technique in use to repair cartilage lesions [27, 169].
Mosaicplasty is a technique that has been introduced for smaller osteo- chondral defects [72, 73]. In mosaicplasty, osteochondral plugs are taken from the less weight-bearing peripheral joint surface and transferred to the corresponding holes in the lesion site. In the space between the plugs, fibrous cartilage takes place.Fresh osteochondral allograftscan be used to re- place large areas of bone and cartilage, and their use is the recommended treatment for young, active patients with large, traumatic osteochondral defects [67]. These grafts are difficult to obtain, which limits their use.
More than ten years ago,soft tissue grafts, such as periosteal or perichon- drial grafts were under intensive examination, but the results have been disappointing, and they are not in widespread clinical use [78, 139, 154].
Cell transplantationswere introduced after the observation that the chond- rocytes around the lesion were not able to move into the repaired area.
Therefore, it was necessary to bring extra cells to produce matrix in the damaged area. Earlier, cultured autologous chondrocytes have been suc- cesfully isolated and transplanted in cartilage defects in rabbits [61].
The method most closely studied in this thesis work,autologous chond- rocyte transplantation(ACT) (also known as Autologous Chondrocyte Im-
plantation (ACI)), was first used in human patients by Brittberget al.[24].
In the study of Brittberget al., a total of 23 patients with deep cartilage de- fects underwent ACT surgery, where autologous chondrocytes were cul- tured under laboratory conditions and injected into the cartilage lesion, covered with a periosteal flap. Three years (36 months) after the operation, 15 out of 16 femoral transplants and 1 out of 7 patellar transplants showed the appearance of hyaline cartilage. Subsequently, ACT has been studi- ed and is now widely accepted as a beneficial cartilage repair technique [57, 119, 146, 147, 157, 184]. It is noted that chondrocytes can produce collagen type II in cell cultures, and biopsies have revealed hyaline–like cartilage or hyaline–fibrous cartilage in cartilage grafts [11, 76, 155].
C
HAPTERIII
MRI of articular cartilage
3.1 Nuclear magnetic resonance (NMR)
Nuclear magnetic resonance (NMR) is based on the interaction of a mag- netic nucleus and its spinswith an external magnetic field, B0. The focus in this work is on hydrogen (1H), also referred to as the proton (since hyd- rogen nuclei contains a single proton). The spin of the nucleus is nonzero when the number of protonsZ and/or the number on neutrons is odd.
The magnetic dipole moment,µ, has the direction of theB0field, giving the direction for the spinning motion of the nuclei. The nucleus can have 2s+ 1energy stages:
E =−msγ~B0 =−ms~ω0, (3.1) wherems=−s, −s+ 1, . . . , s−1, s and ~is Dirac’s constant. 1H has two possible energy levels: parallel (+1/2) or anti-parallel (−1/2) state with respect to the static field. The uneven distribution of the proton popula- tions is given by the Bolzmann equation
N−1/2
N+1/2
=e−∆E/kT =e−~ω0/kT, (3.2) whereN is the number of nuclei in the spin orientation, ∆E is the energy difference between the states,T is the temperature andk is Boltzmann’s constant (1.38x10−23J/K). As the spin of the hydrogen nucleus is (±1/2), it has only one quantized energy describing the transition between the states. Since there is a large number of a hydrogen nuclei in biological tissues, this is a perfect match for biological and medical MR applications.
This large number of hydrogen nuclei in the tissue gives rise to the net magnetizationM0 along theB0.
In a staticB0 field, nuclei will precess in the direction of the magnetic field at Larmor frequency given by
ω0 =γB0, (3.3)
whereγis a constant called the gyromagnetic ratio. For example, in water, the hydrogen proton has a value ofγ=2.68x108rad/s/T or 42.6MHz/T).
3.2 Relaxation
An additional quantum of energy can change the direction of the spin state of the nucleus away from the direction of the z–axis, provided that it is introduced with a frequency matching the Larmor frequency of the given nuclei. This is called a rf–pulse that induces a magnetic field,B1. Applying B1for a duration twill tilt the net magnetization90◦ (π/2 pulse),i.e. into the xy–plane, leading to zero magnetization in the direction of the z axis.
A pulse at the same power but duration 2t will rotate the magnetization by180◦(πpulse),i.e.it will invert the orientation of spin populations. The field in the xy–plane is described as
B1 =B1(cosω0txˆ−sinω0ty),ˆ (3.4) inducing a time dependent behaviour of the net magnetization:
dM
dt =γM×Bext, (3.5)
whereBext =B0+B1. The different components ofM are given as dMx
dt =γ(MyB0+MzB1sinω0t), (3.6) dMy
dt =γ(MzB1cosω0t−MxB0), (3.7) dMz
dt =γ(MxB1sinω0t+MyB1cosω0t). (3.8) After the rf–pulse, the excited spins will immediately start to arrange to- wards the equilibrium, a phenomenon called relaxation. The change in magnetization induces a current into the rf coil, producing a signal known as the free induction decay (FID). Since this is time dependent, the fre- quency distribution can be revealed using Fourier transform (FT). The re- covery of magnetization is represented by the Bloch equations [19]
dMxy
dt =−Mxy
T2 , (3.9)
3.2 Relaxation 31 dMz
dt = M0−Mz
T1 . (3.10)
The recovery in the z–axis direction is called longitudinal or spin–lattice relaxation,T1, and the decay in the xy–plane is called transversal or spin–
spin relaxation,T2.
3.2.1 Spin–lattice relaxation (T1)
Protons are in thermal contact with the lattice of their nearby atoms, and each proton will experience different magnetic field variations. In addi- tion, the local magnetic fields created by rotation, translation and vibra- tion of these atoms affect the field variations. After the 90◦ rf–pulse, the longitudinal magnetization evolves in an exponential manner, until the z–
component of net magnetization is recovered. This can be derived from equation 3.10:
Mz(t) =Mz(0)e−t/T1 +M0(1−e−t/T1). (3.11) The relaxation occurs when a component of the fluctuation frequency matc- hes the nuclear Larmor frequency and stimulates a spin flip. The rate of field fluctuations is characterized by thecorrelation timeτc. The relationship betweenT1relaxation rate and the frequency distribution of the molecular motion can be described as [26, 54]:
1 T1
∝Bxy2 τc
1 +ω02τc2, (3.12) whereω0=2πf0 is the resonance frequency. The relaxation is most efficient whenτc=1/ω0.
Inversion recovery is a technique for producing T1contrast, consisting of a combination of two rf–pulses. First, a π–pulse is introduced which inverts the equilibrium magnetization. Longitudinal magnetization now starts to increase and no transversal magnetization is created. The net magnetization vector will eventually return back to +z axis at a rate de- termined by T1. Since the magnetization on z axis is not detectable, a π/2–pulse is introduced to tip the longitudinal magnetization back to the xy–plane. Following theπ/2–pulse, the signal is given as [71]
Mz(T I) = 0 (3.13)
Mxy(T I) =M0(1−2e−T I/T1), (3.14) whereTIis the time between the inversion pulse and the 90–degree pulse, i.e. inversion time, and M0 is the net magnetization. The selection of TI
defines the amount of T1–weighting in the registered signal. When this experiment is repeated with several TI values,T1 can be determined us- ing equation 3.14. The factor 2 in equation 3.14 assumes that the magne- tization is at full inversion. This factor can also be fitted to account for inaccuracies in the inversion.
Previously, native T1 relaxation time of cartilage has been shown to correlate with biomechanical parameters, degeneration stage, proteogly- can depletion [99, 135, 138, 175, 183] and T1–weighted imaging has been used to evaluate repaired cartilage and cartilage lesions [35, 48, 49, 188].
In addition,T1 has been claimed to be able to monitor the biophysical pro- perties of engineered cartilage [120]. However, the relationship between T1and cartilage composition is not fully understood. It has been reported to be relatively constant throughout the tissue depths [122, 197], and to be isotropic,i.e. showing no orientational dependence [77].
3.2.2 Spin–spin relaxation (T2)
In addition to the applied field, spins experience differences in the local field due to their mutual presence. This leads to different local precession frequencies. Energy exchange with the lattice is not involved. Due to the variations on the local magnetic field, the individual magnetic moments will gradually lose their phase coherence, leading to a dephasing of the net magnetization vector. This leads to a signal decay, known as spin–spin relaxation. Solving the equation 3.9, transverse relaxation is given by
Mxy(t) =Mxy(0)e−t/T2. (3.15) The relationship between the relaxation rate 1/T2 (=R2) and spectral den- sity is described as
1 T2
∝Bz2τc. (3.16)
As well as the dephasing of individual spins, there is also additional de- phasing caused by field inhomogeneities. The total relaxation time (T2∗) is a consequence of these terms given by
1 T2∗ = 1
T2
+ 1
T20, (3.17)
where T20 is the transverse relaxation due to the presence of field homo- geneities causing a signal decay. This can be compensated with a method called the spin echo technique. This contains two rf–pulses, a π/2–pulse followed by a refocusing π–pulse. The first pulse tips the magnetization
3.2 Relaxation 33 into the xy–plane and the spins begin to dephase. After a timeτ, a second pulse reverses the spins in such a way that the spins earlier experiencing a positive phase now experience a negative phase, and vice versa. After theπ–pulse, the spins begin to rephase creating an echo. The time from the90◦pulse to the maximum intensity of the spin–echo is called theecho time,TE=2τ. The detected magnetization is given by
Mxy(T E) =Mxy(0)e−T E/T2. (3.18) TheT2relaxation time constant can then be determined from the signal in- tensity measurements with variableTE–times. When only one refocusing pulse is applied after excitation, the experiment is called asingle echo spin echosequence.
Rubenstein et al. reported the strong orientation dependence of T2– weighted imaging in cartilage due to the oriented collagen structure [158].
Subsequently,T2–weighted imaging has been often used to visualize the network arrangement within the cartilage layers, due to the magic angle effect [59, 65, 134, 198]. Depending on the cartilage orientation inB0, the dipolar interaction within zones become altered, thus affecting T2. This interaction is at its minimum at54.7◦,i.e. at the magic angle. This appears with high signal intensity inT2–weighted images.
T2imaging has been used in knee MRI studies, concerning cartilage le- sions whereT2values of the focal cartilage defects were found to be higher than values in the adjacent cartilage [102]. Osteoarthritis studies noted thatT2mapping of articular cartilage may reveal early cartilage lesions not visible with standard clinical MRI and may be useful in quantifying early OA related changes [74, 167]. ACT cartilage repair techniques have been studied with T2 relaxation time by White et al. [193]. They showed that qualitative and quantitative T2 mapping helped to differentiate hyaline cartilage from reparative fibrocartilage after cartilage repair.T2–weighted imaging was reported to be unable to predict ACI graft histological fea- tures [174], whereas another study claims MRI to be a useful non–invasive tool for evaluating the morphologic status of osteochondral plug transfers [182].
Proton density orT2relaxation time measurements have also been spe- culated to reflect the water content in the tissue [106, 109, 162]. T2 has also been proposed to depend on the level of PGs [150, 189]. T2 has been related to the mechanical properties of articular cartilage, since it revealed a significant correlation with Young’s modulus and dynamic modulus [99, 135, 138, 189]. T2 imaging has been also shown to be able to differentiate degeneration [138, 141], maturity [140], morphology [200] and topograp- hical variations [135, 187] of cartilage samples.
3.3 dGEMRIC
A contrast medium that is frequently used in MRI is a gadolinium comp- lex with diethylenetriamine pentaacetic acid (Gd-DTPA2−), supplied in the form of dimethylglucamine salt. Free Gd3+ is very rare and toxic, and therefore it is necessary to bind it with a high stability constant chelate.
The gadolinium atom carries seven unpaired electrons and hence is strong- ly paramagnetic, shortening the relaxation times [47]. Other important paramagnetic ions are chromium ( Cr2+), manganese (Mn2+ and Mn3+) and iron (Fe2+), often embedded in chelates when used in MR studies.
The contrast agent concentration can be presented as [Gd−DT P A2−] = 1
R( 1 T1Gd
− 1 T1
), (3.19)
where R is the relaxivity of Gd–DTPA in (mM−1s−1)), often expected to be the value of saline solution [38], T1 and T1Gd are the relaxation times without and with the contrast agent, respectively.
Bashir et al. measured the human cartilage GAG concentration with the gadolinium enhanced MRI of articular cartilage (dGEMRIC) technique [13, 14]. Since GAGs have negatively charged side groups, Gd-DTPA2− ions will be distributed in cartilage, reflecting the local GAG concentra- tion, with higher concentrations in those areas with depleted GAGs and vice versa. The spatial contrast agent concentration is inversely related to GAGs, the main source of tissue FCD. Bashir et al. noted the connection between Gd-DTPA2−and FCD [13, 14]:
F CD = 2[Na+]b(
s[Gd−DT P A2−]t [Gd−DT P A2−]b
−
s[Gd−DT P A2−]b [Gd−DT P A2−]t
), (3.20) wheretandbrefer to tissue and bath, respectively.
The relaxivity differs between tissues and magnetic field strengths [38, 165]. The T1Gd relaxation time is related approximately linearly with the GAG content of cartilage [13, 14]. Subsequently, dGEMRIC became a widely accepted and used method to measure cartilage GAGsin vivo[12, 107, 115, 156, 173, 179] and in vitro [9, 99, 125, 133, 179]. dGEMRIC has been shown to predict the compressive stiffness of articular cartilage in vitro[99, 135, 138, 159, 189]. Protocol issues have been published [29] and the technique has been used to evaluate the healing process and GAG con- tent in ACT patients [55, 184]. The sensitivity of the dGEMRIC technique has been shown to be good in hip dysplasia and in OA studies it can iden- tify poor candidates for a pelvic osteotomy and dGEMRIC values have correlated with pain and the severity of the dysplasia [33, 91, 172].
3.4 Other quantitative MRI methods 35 Limitation in the accuracy of the dGEMRIC method was raised in a study where relaxivity was shown to be dependent on the macromolecular content [168]. However, cartilage was not studied.
Recently, Nieminenet al. suggested that it might be possible to com- bine the imaging sessions from T2 and dGEMRIC [132]. Until now T2
and dGEMRIC are being imaged in separate imaging sessions since Gd- DTPA2− may affect native T2 by offering an additional relaxation mecha- nism. The effect of Gd-DTPA2− was greater with high concentration and long T2s while the deep tissue which have relatively low T2s and [Gd- DTPA2−]values was not significantly altered by Gd-DTPA2−[132]. Finally, it has been shown that it is possible to obtain accurate morphological mea- surements of cartilage in the presence of Gd-DTPA2−and that morpholo- gical and dGEMRIC measurements may be combined in a single imaging session [43].
3.4 Other quantitative MRI methods
Spin–lattice relaxation in the rotating frame,T1ρrelaxation, provides infor- mation about macromolecules with slow rotational motions. In this spin–
lock method, spins are ’locked’ in the xy–plane by applying a continu- ous rf–pulse. Magnetic moments are then precessed around the spin–lock field. During a spin–lock pulse, the magnetization relaxes towards equi- librium with the relaxation constantT1ρ.T1ρhas been found to be sensitive to the cartilage PGs [1, 40, 41, 105, 144, 151, 152]. Relaxation mechanisms in the rotating frame in cartilage have also been investigated, revealing thatT1ρ is also dependent on the collagen orientation of the cartilage due to residual dipolar coupling [124].
Magnetization transfer, MT, involves the exchange of magnetization bet- ween bound (i.e. immobilized or adsorbed) water and protein protons.
In a MT experiment, a long rf pulse is applied at the off–resonance fre- quency from the bulk water resonance, saturating the proton magnetiza- tion. When the components (i.e. water or protein protons) are saturated, the exchange of magnetization occurs until a steady state is achieved. A reduction of the signal intensity of the bulk water is then observed. In car- tilage, collagen plays the main role in MT, since PGs make only a minor contribution to MT [63, 100, 101, 160, 185].
Diffusion is the random motion of the water molecules in the tissue.
Diffusion tensor imaging(DTI) experiment is done by applying diffusion–- sensitizing gradients and registering diffusion–related signal attenuation.
The self–diffusion of the water protons characterised by a 3x3 tensor, de- scribing both the magnitude and the direction of the diffusion in a 3–
dimensional space [15]. It has been shown that the DTI method can be used to measure diffusion anisotropy in human cartilage and that the di- rection of the maximum diffusion reflects the alignment of collagen fibers, i.e. motional anisotropy of water is a consequence of the attraction or binding water molecules by collagen [50]. Diffusion constants may also reflect structural degradation of the cartilage matrix. Apparent diffusion constant (ADC) has been shown to be sensitive to proteoglycan depletion [175] and may reflect the structural degradation of the cartilage matrix [123]. DTI has been studied in enzymatically degenerated bovine [116]
and human [34] cartilage, indicating that GAG loss slightly increases the diffusion anisotropy and the ADC in the cartilage. However, no changes were noted in fractional anisotropy [34, 116].
The negatively charged side groups of GAGs attract sodium ions arou- nd them, assuring electroneutrality in the tissue. The early stage of OA is primarily associated with a loss of PGs, which leads to a decrease in the sodium concentration. Based on this, sodium has been used to reflect the PG depletion in the cartilage [21, 31, 108]. Shapiro et al. showed that sodium assessed accurately the FCD in articular cartilage [161]. Sodium MR imaging has also shown to represent a potential method for use as a quantitative diagnostic tool to measure changes in proteoglycan content in early-stage osteoarthritis [192].
Collagen fibers attract water molecules by inducing motional aniso- tropy. The signal of these water molecules can be effectively detected by the 2H spectroscopic imaging technique that is based on the distribution of the2H quadrupolar splitting and further the spatial orientation of col- lagen fibers [88, 131, 163]. 2H spectroscopic imaging has been shown to be sensitive to the order and density of the collagen fibers in pig articu- lar cartilage from birth to maturity [89]. The effect of load applied to the cartilage-bone plug has been monitored and the orientation and the deg- ree of order of the collagen fibers at each spatial location on a cartilage plug has been estimated [164].
Since thinning in cartilage thickness is involved in OA, in vivo carti- lage thickness and volume have been a subject of interest for clinicians.
Females with a higher incidence of knee osteoarthritis (OA) than males have thinner cartilage and smaller joint surfaces, even after adjustment for height and body weight [142]. Nonetheless, it has been shown that thin cartilage does not predispose to OA [80]. In addition, the factors stimula- ting bone and cartilage growth may differ between the sexes [143]. The fat suppression gradient echo imaging offers the possibility to study carti- lage morphometry. Building three–dimensional virtual computer presen- tations, it is possible to improve accuracy compared with two–dimensional
3.4 Other quantitative MRI methods 37 plane images [20, 22, 44, 45, 47, 56, 94, 104]. 3D imaging has also been suc- cessfully used to shorten the examination time in articular cartilage ima- ging [176], for monitoring the healing process of the lesion after surgical cartilage repair [178] and, together with dGEMRIC, to evaluate the relative glycosaminoglycan content of repair tissue after matrix–associated auto- logous chondrocyte transplantation (MACT) [178, 180], which is a three–
dimensional biomaterial scaffold used as a carrier for chondrocytes. 3D imaging has also been used successfully in a fast semi–automated software method to segment the cartilage in knee MRI [39]. 3D MRI was shown to measure accurately and reliably small changes in cartilage volumeex vivo [85].
Normal tendons, ligaments and uncalcified fibrocartilage produce little or no signal and they appear dark with all pulse sequences caused by their shortT2s. With ultrashort TE (UTE), it is possible to identify in a specific manner, the calcified cartilage and uncalcified fibrocartilage [16, 181].
C
HAPTERIV
Aims of the present study
Several quantitative magnetic resonance imaging techniques, based on1H NMR relaxation properties, have been developed for the characterization of structure and composition of articular cartilage. The present study has applied qMRI techniques in an attempt to elucidate variations in proper- ties of native and repaired cartilage in the animal and human joint, and in this way to evaluate the clinical applicability of quantitative MRI tech- niques. The aims of the presentin vitroandin vivostudies were
1. to study the topographical variations of the cartilage MRI properties in the human knee joint and to relateT2relaxation time and dGEM- RIC imaging techniques with the structural and biomechanical pro- perties of human cartilage;
2. to study whetherT2and dGEMRIC imaging techniques can be mer- ged into one imaging session and be able to produce reliable quan- titative information on the collagen structure of cartilage using T2
relaxation time measurements in the presence of Gd-DTPA2−;
3. to study whether nativeT1relaxation time could serve as a biomarker to characterize the water content of articular cartilage;
4. to investigate the importance of combining T2 relaxation time and dGEMRIC techniques to monitor the regeneration of cartilage tissue after ACT surgery.
C
HAPTERV
Materials and methods
The present thesis consists of four independent studies (I-IV), threein vitro studies with cadaver human or bovine samples and onein vivo study in ACT patients. A summary of the methods used is presented in table 5.1.
5.1 In vitro experiments
5.1.1 Sample preparation CADAVER HUMAN SAMPLES
In studies I and II, left knees of human cadavers (n= 13, age 20–80 years) were obtained from Jyväskylä Central Hospital, Jyväskylä, Finland, with permission from the national authority (National Authority of Medicole- gal Affairs, Helsinki, Finland, permission 1781/32/200/01). Knees were frozen post–mortem and, after thawing, full–thickness cartilage–bone cy- linders (diameter = 16 mm, n = 78) with subchondral bone were drilled from nonarthritic knees at six anatomical locations: the latero–proximal patella (PAT), medial/lateral condyles of the femur (FMC/FLC), medial/- lateral tibial plateaus (TMP/TLP) and the femoral groove (FG) (figure 5.1).
One patellar sample was excluded because of cartilage degeneration. The samples (n= 77) were frozen at−20◦Cafter immersing them in phosphate–
buffered saline (PBS; Euroclone Ltd., Paignton–Devon, UK) containing en- zyme inhibitors (5 mM ethylenediaminetetraacetic acid (EDTA) (Merck, Damstadt, Germany) and 5 mM benzamidine HCl (Sigma, St. Louis, MO)).
Before the measurements, the osteochondral plugs were thawed, and smal- ler full–thickness cartilage disks (diameter = 4.0 mm) without subchondral bone were prepared with the use of a biopsy punch and a razor blade. Six anatomical sites were included in study I, whereas in study II knees with- out patellar samples were examined (n= 65).
